Prevención no represiva

Obtaining and evaluating the particle size distributions of hydrometeors at the Mario Zucchelli research station is one on the aims of this thesis. To reach this goal the Parsivel disdrometer data, collected during the austral summers 2016-2017 and 2017-2018, were processed and evaluated. This represent the rst analysis in which microphysical data of two seasons from MZS have been used.

Both PSD (Figures 4.11 and 4.14) highlight the presence of very small ice particles during the summer periods. This feature is consistent with the dimensions measured at other Antarctic locations (Lachlan-Cope et al. 2001; Gay et al. 2002; Walden et al. 2003), reporting values between 0.03 and 0.6 mm. It must be underlined that most of these observations are referred to research stations very far from the sea. On the Antarctic coast, due to the inuence of the ocean on the atmospheric moisture amount, the parti- cle dimensions are expected to be slightly larger, and this proved to be in reasonable agreement with the microphysical results found at MZS. More- over, Gorodetskaya et al. (2015) reported that photographic observations of the fallen snow crystals at the Princess Elisabeth research station captured a maximum size of 0.5-0.8 mm, whereas Souverijns et al. (2017) indicated that the snow particles over the PE station have a median diameter around 0.7 mm. Considering the position of the PE station, a few hundred kilome- ters away from the coastline, both the maximum dimension and the median

diameter of the hydrometeors found at MZS are consistent with the PE ob- servations.

Analyzing the spectrographs (Figures 4.13 and 4.16), the inuence of high-speed wind in the Parsivel retrieval cannot be neglected. As stressed in section 3.1.2, the disdrometer betrays a lack of accuracy during windy days. The spectrograph of the rst observation period (2016-2017) exhibits a very high number of small particles, coupled with an unrealistic fall speed. This can be considered a peculiar signature of recurring high-speed wind conditions at MZS. This well-known drawback of the instrument was deeply investigated by Friedrich et al. (2013a,b). They found an artifact in the Parsivel observations represented by an unusual large number concentration of hydrometeors with large diameters and low fall speed in windy conditions. This characteristic feature was not found in the Parsivel measurements at MZS, where small particles with unrealistic high speed were clearly detected when strong winds blew. This disdrometer limitation should be born in mind when interpreting the Parsivel data, and further mitigation measures must be devised and applied in order to improve the retrieval of the disdrometer. However, the high number of small hydrometeors captured by the in- strument in windy conditions could also be caused by other factors, as well as due to the intrinsic limitation in small particles detection of the Par- sivel (Battaglia et al., 2010). Blowing snow contributes to the near ground snow ux, and is particularly eective in recirculating small and light parti- cles (e.g. Mann et al. 2000; Gordon and Taylor 2009). Gorodetskaya et al. (2015) report a threshold of 8 m/s, as wind speed needed for lifting snow from the ground, a value often exceeded at MZS. Moreover, as evaluated by Vardiman (1978) and supposed by Grazioli et al. (2017b), a contribution to small particles may be the fragmentation of aggregates, in particular at the

lower levels of the atmosphere, due to the mechanical breakup caused by strong wind and turbulence. Both these eects can be key contributions to the high concentrations and to the unrealistic velocities of the small particles observed.

Finally, the accumulated snowfall amount was calculated for each Antarc- tic expedition using the Parsivel data in spite of its limitations. The formula presented by Huang et al. (2010) was used, even though it was derived for 2DVD disdrometer that measures two sizes of each ice particle. The snow- fall rate, calculated through such formula, proves to be very sensitive to the density-size relation used. Moreover, the observation periods are limited to just a few months. For these reasons, it is therefore dicult to compare the results obtained in this analysis with the other annual estimates of accumu- lated snowfall at MZS reported in the literature, although, at rst glance, they seem to be substantially higher, conrming that the quantication of precipitation in the coastal regions of Antarctica remains a dicult task.

5.1.3 Hydrometeor characterization

The simulated equivalent radar reectivity factor was calculated through the formula 3.27, using the outcomes of the four backscattering cross section com- putations and the particle size distributions. Comparing the simulated Ze

values with the simultaneous actual reectivity measured by the Micro Rain Radar, the features of the precipitation at MZS can be deduced. Despite of the uncertainties of the Parsivel measurements, considering the various approximations made in the DDA database processing and in the T-Matrix computations, as well as the distance between the lowest measuring range of MRR (300 m) and the Parsivel instrument at the ground, the compar- isons result in an unexpected agreement, albeit in 6 precipitation events the

methodology failed. For the aggregate-like events the mean value of the cor- relation coecient is 0.68, well over the xed threshold, whereas the RMSE mean is slightly more than 5 dBz. The precipitation days classied as hav- ing pristine-like features exhibit a mean Pearson index of 0.69 together with 4.9 dBz as RMSE average. These indexes show that the methodology works generally well.

Moreover, two comments on the scattering methods can be derived from these comparisons. First, the DDA aggregate and pristine databases repro- duce satisfactorily the scattering properties of ice particles at MZS, second, the T-Matrix simulations again appear to be very sensitive to the size rela- tions. In fact, a good correspondence between the T-Matrix method and the DDA-pristine was found when the CANTMAT code was initialized with the relationships obtained from the NASA database, even though such relation- ships have been derived only from the aggregate database. This conrms that the T-Matrix seems to be having trouble in dealing with aggregate particles. Instead, when the literature relations were used in the CANTMAT code, the simulations result in a systematic overestimation of the actual reectivity values.

The classication of the events reveals that ice particles frequently exhibit an aggregate- or pristine-like form at MZS. The only other categorization, in the coastal regions of Antarctica, was carried out by Grazioli et al. (2017b) that classied the hydrometeors exploiting the polarimetric data of the MX- Pol radar at the DDU research station. They observed, at 400 m above the ground, a proportion of three particle types with about 10% of rimed snowakes, 40% of aggregates and 50% of crystals, albeit at the ground the majority of the hydrometeors (54%) was identied as small particles, and 20% was by aggregates. This research can in part conrm the results found

in this thesis inasmuch the presence of tiny solid hydrometeors and a sub- stantial number of aggregates have been recognized. However, as opposed to Grazioli et al. (2017b), the methodology applied in this dissertation does not allow to discriminate multiple types of hydrometeors in the same event, thus loosing the ne variability in the habits of particles during precipitation.

Analyzing the meteorological observations at the research station, a high correlation was found between the occurrences of aggregate-like precipitation and high-speed wind conditions. Going in depth, we found that aggregate events always take place together with barrier wind, namely when southwest- erly ow, almost parallel to the coast-line, blows at ground level. Thus, we can speculate that the aggregate events are intrinsically related to the pres- ence of a large low pressure system oshore over the Ross Sea, that pushes air masses from the ocean towards the steep coast, increasing the moisture and promoting aggregates formation. Furthermore, the turbulence and the high-speed wind could broke the aggregates at lower levels, as mentioned before, explaining the high concentration of small particles captured by the disdrometer. The pristine-like episodes are found to occur mainly in con- junction with a low-speed wind with variable direction. Moreover, they seem to be related to a meteorological pattern in which the air masses originating from the inner part of Antarctica (where single crystal precipitation have re- peatedly been observed (e.g. Lachlan-Cope et al. 2001; Walden et al. 2003) likely due to the extremely low humidity level) to reach the Mario Zucchelli station.

On the other hand, the last two points can also be interpreted as due to the diculty of the Parsivel in ice particles detection, giving rise to obser- vations biased because of the wind. This is the reason why another type of ground instrument (e.g. 2DVD, MASC) or photographs of the falling parti-

cles could be helpful to completely demonstrate the validity of the method- ology presented in this work. Actually, a photograph of the hydrometeors at

Fig. 5.1: Picture of the fallen hydrometeors at the Mario Zucchelli research station on 2 December 2016.

MZS was taken on 2 December 2016. That day was classied as an aggregate- like episode (correlation coecient=0.65, RMSE=5.5 dBz) by the method presented in this thesis. The hydrometeors depicted in the picture seem to be snowakes aggregates having dimension around 1 mm (Roberto et al., 2018), conrming the validity of the methodology.

It should also be stressed that the importance of a correct classication of the hydrometeors lies in the possibility of using the appropriate density- size relation in each precipitation episode. This relationship represents a key term in the snowfall rate formula, and therefore in the calculation of the total snowfall accumulation at the ground, that is the fundamental input for establishing the hydrologic cycle of Antarctica.

Finally, as well demonstrated by the unclassied event presented in sec- tion 4, comparing simulated Ze values from Parsivel observations with the

reectivity measured by the Micro Rain Radar, the blowing snow episodes can be detected. In fact, the height above the ground of the lowest MRR gate (300 m) can be considered not aected by this phenomenon (e.g. Gordon and Taylor 2009; Scarchilli et al. 2010), which rarely exceeds 200 meters of vertical development (Grazioli et al., 2017b). Hence, according to Grazioli et al. (2017b), the synergetic use of remote sensing and in-situ measurements can overcome in part the problem of the phantom precipitation, detecting, and therefore removing, the contribution of the blowing snow to the total accumulation. The main limit of this treatment seems to be that it can be applied only to pure blowing snow episodes, hence not when precipita- tion and blowing snow occur together. Actually, considering the dierence between simulated and true reectivity values an estimate of the contribu- tion of blowing snow could be derived if an appropriate Ze-S relationship is

used. In the same way also the low-level sublimation episodes, due to the dry katabatic winds, can be detected when the reectivity at 300 meters exceeds considerably the simulated reectivity at the ground. According to Grazioli et al. (2017a), the sublimation of precipitation accounts up to 35% reduction of total snowfall on the margins of East Antarctica, thus posing a serious problem to the measurements of snowfall accumulations as well as to the satellite-based estimations of precipitation.